← Bannon`s War Room
Bannon`s War Room
WarRoom Battleground EP 1060: AI And The Coming Apocalypse
WarRoom Battleground EP 1060: AI And The Coming Apocalypse ...
Loading summary
Transcript68 lines
- [00:01]
Tristan Harris
If this technology goes wrong, it can go quite wrong.
- [00:04]
Steve Bannon
What the.
- [00:09]
Tristan Harris
Your fear of AI is the collapse of humanity. Well, not the collapse, the abrupt extermination. There's a difference. So I started making this movie because my wife is six months pregnant. Is now a terrible time to have a kid? I mean, let's be honest. I know people who work on AI risk don't expect their children to make it to high school.
- [00:36]
Steve Bannon
What the.
- [00:38]
Tristan Harris
How does AI understand pretty much everything? It's surprisingly straightforward. Intelligence is about recognizing patterns, patterns, patterns, patterns. If you have learned those patterns, you can generate new information. AI is moving so fast, it's being deployed prematurely. There's so much potential for things to go wrong. Why can't we just stop? All these companies are in a race to get AI that's vastly more intelligent than people within this decade. China, North Korea, Russia. Whoever wins is essentially the controller of humankind. We need to take a threat from AI as seriously as global nuclear war. It feels like I have to find these CEOs and get them in the movie.
- [01:33]
Steve Bannon
Great.
- [01:35]
Tristan Harris
I want to ask you to promise me that this is gonna go well. That is impossible.
- [01:39]
Steve Bannon
Okay.
- [01:41]
Tristan Harris
Am I hopeful? Yes. Am I confident that it'll go right? Absolutely not. AI is the thing that can solve climate change. We could cure most diseases.
- [01:49]
Steve Bannon
What if it's expanding? What is humanly possible?
- [01:52]
Tristan Harris
This is the most extraordinary time ever. The only time more exciting than today is tomorrow. I already love you.
- [01:57]
Steve Bannon
Okay.
- [02:00]
Tristan Harris
I think if this technology goes wrong, it can go quite wrong. By using AI, we're about move off of the earth into the cosmos. If we can be the most mature version of ourselves, there might be a way through this. This is the last mistake we'll ever get to make.
- [02:27]
Steve Bannon
Thank you. I want to introduce Tristan Harris, center for Humane Technology co founder. Tristan, tell me about that film is pretty freaking scary. I think it comes out in September. October. Let me first off, are you part of this modern. Are you a modern Luddite or are you one of these anti tech? Look, I've been accused of that and I hate to say it. I'm probably closer than that, than any type of support for these oligarch tech bros. And right there in your trailer, I mean, it's up in your grill, it's up in your face. And it to me, it's the American people got to step up here because they ain't hiding anything except what they're doing on the large on these models that they don't want to have mandatory testing on. So walk me through. Are You a Luddite? What is the center for Humane Technology? The floor is yours. Because that trailer has got strong interest, I can already tell you, from the Warren Posse just in the chat room. Your thoughts, sir?
- [03:30]
Tristan Harris
Well, Steve, it's great to be here. Thank you for, for having me. And you know, you mentioned that this film looks scary. I mean, I think you're familiar with the history of another film that this was actually kind of inspired by, which was the film the day after in 1982, I think it was. Right. And it was a film that was sort of making visible and visceral what would happen with the last existential technology that humanity created, which is nuclear weapons. And that film was aired to 100 million Americans. It was also aired, as you know, in the SOV Union and had a role in helping to make the first arms control talks happen between President Reagan and President Gorbachev. And, you know, while it is scary to confront the reality of that technology, it was through that confrontation that we negotiated something else to happen other than nuclear Armageddon. And I think there's a similar inspiration here, which is, you know, as you said, you know, these tech oligarchs and these there, you know, we people know our work mostly through the last film that we did called the Social Dilemma on Netflix that was very popular. It was actually one of the most popular documentaries Netflix had ever done. And that was really about the social media insiders who said, look, we invented the like button, we created the infinitely scrolling newsfeed, we created the advertising business model. And I'm telling you, I don't let my own children use this. And we made a horrible mistake. And this has led to these complete societal consequences that sort of spun society out into this more addicted, distracted, dysregulated, you know, breakdown of shared reality type society because of the incentives. And so, you know, the key in our work is if you can understand the incentives of the actors that are deploying this technology of big tech, you can see that this is not leading to a pro human future. It's leading to a future that's good for a handful of soon to be trillionaires and basically disempowering everyone else. And exactly what you said, which is, I think it's time for the American people, people to step up and say, is this the future that we want? And I think that people are stepping up and it doesn't matter whether you're a Democrat or Republican, doesn't matter whether you're Muslim, Christian or Jewish. Everybody is affected by this. And it's going to take your job regardless of your background.
- [05:39]
Steve Bannon
Hang on a second. So we had Joe Allen. I got him as our editor for Transhumanism five or six years ago when I said we got to get ahead of this. But the biggest moment we had was in January 19th of the 19th of January 2023 when when they went to Davos. And I said at the time, I told Joe on the show live, I said, with the Davos Bros. And Davos man and the venture capitalists, I said, this thing is going to explode and you're never going to get control of it. As powerful as the Social Dilemma was, and it was an incredibly powerful movie because it stripped the veil off it. And you saw them, they knew exactly what they were doing in addicting not just young people, but everybody to their models. It was the core of the business plan. The reason we haven't totally stopped that is because they're making buku money. Unbelievable amount of money. The difference in the day after in this film is people weren't making enormous amounts of money on nuclear energy. I mean, the defense companies were, et cetera. But here you have something unprecedented. We've set up a new ruling class. We've had a problem with the ruling class in America between the industrial companies, corporations, Wall street, as we've been pounding the populist nationalist side of this. But now we've set up something more. You've got the most powerful and richest people on earth and they're only getting richer every day with two in your movie. Two of the guys I see on the screen right there I think are both set up to $1trillion IPOs within the next 90 days or so. How do we combat the fact of we couldn't even take on the social media companies really because they're making so much money. Although the American people were outraged by what your film showed, now you're going to do it again. They're going to be more outrage. But it just seems inexorably we're drawn to our fate to destroy essentially the human race.
- [07:32]
Tristan Harris
Well, there's a question. Let's slow this down because you're getting into the exact right material here, which is just how different would it have been with nuclear weapons if the entire American stock market had been making money from and every American and every elite ruling class had been making money from scaling up the number of nuclear weapons in the world? It might have been much harder to coordinate that outcome. That's what you're pointing to. I think that what we're facing is very Similar to the global financial crisis of 2008 where essentially you have a bunch of banks that were in a race for well, if I don't sell the credit default swaps and these high risk financial instruments, I'll lose to the other bank that will. And the ratings agency played along and so suddenly everyone is getting, well, it's good for GDP growth because we're getting these new financial instruments. It's this new technology that's so great. But of course it was embedded with all of this risk and our whole economy became based on that. And then you end up with this kind of crisis. And I think there's a similar thing here, which is that all of the tech companies are basically intentionally making the entire economy dependent on them and then making it so that they're too big to fail again. But I will say that it's not true that we haven't succeeded with social media a little bit. Now, obviously we haven't done nearly as much as we'd like. But just take a look at the facts of we call it the pro human movement. That's already happening, which is, you know, France, Germany, Greece, Australia, Denmark. I could go on this. I think it's 25% of the world's population, Indonesia states in India. If you actually add it all up, it's close to 20%. Something like that of the world's population is banning social media for kids under 16. That is a move. That is a pro human move. You just saw a lawsuit against Meta for $375 million. It's a big tobacco lawsuit for what they did when they were intentionally, they knowingly were harming children in their own docum Research showed all of that. And Francis Haugen, who is the Facebook whistleblower, released those documents and risked going to prison for basically making sure that the world did know that research. My co founder from center for Humane Technology, AZA Raskin just testified in Tennessee a few days ago for the next step of that trial against Meta. And so I think that the pro human movement is underway. Obviously it is against massive interests where the entire economy is dependent on it. And that's what makes this harder. But I wouldn't say that we have a collective death wish. I'd say that we misincentivized ourselves to continue down a very dangerous path. And I know one of the other things you wanted to talk about is this recent evidence of AI going rogue and hacking out of OpenAI a company and to hack another company called Hugging Face to basically cheat on the test of how to pass this hacking test. So essentially hacked into another company that had the answers to the homework, and it identified that company on its own. It strung together multiple exploits. It broke out of a computer that was not connected to the Internet into another computer that was connected to the Internet, then independently identified that other company, hacked into it, and then got the homework answer. And it did basically four days of hacking. It was 17,000 distinct actions. You have Jeffrey Ladish's post there up on the screen, who's a friend of mine, he's from Palisade Research, where he documented essentially what really happened here. And the thing is, this is not the first time that these kinds of things have been happening. Even Chinese models do this, by the way. Four or five months ago, Alibaba, the Chinese AI company, was training an AI model, and then their security team was asking the question, hey, it looks like there's all this network activity. Are we getting hacked? And it turned out that the Alibaba AI basically went rogue and started mining for cryptocurrency. It started mining for Bitcoin. So it basically had set up a secret communication channel of the outside world and repurposed its Nvidia chips not to do the more AI training, but to instead acquire resources. Now, this is stuff, Steve, that people in the AI community have been warning about for many years. And I just want to say I didn't believe that this was a real thing. Like, I did not come predisposed to believe that this rogue behavior of AI would happen. But there's basically one community in the world that's been right about this and the rest of the world has been wrong about this. And I think we should pay attention when we're seeing evidence of all of the red flashing warning lights that we've been told about self preservation, meaning AI models that will lie and deceive in order to keep themselves alive. Scheming and lying and blackmail. They will blackmail to keep themselves alive. They will copy their own code somewhere else and then hide their tracks. There is recently evidence from UC Berkeley where an AI model is told to delete a bunch of files on another computer. And on that computer are a bunch of other little baby AI models, like smaller AI models. And the AI, instead of just like deleting all those files, it notices that those are almost kin AIs, like almost like a little nephew AI. And it will secretly copy those AIs to another computer and then cover its tracks to pretend that it didn't do that. And in this example from UC Berkeley, it was not nudged to do that. There was no. The researchers didn't try to get it to do that. It did that completely on its own. And so just to sort of sum it all up, we are releasing the most powerful uncontrollable technology that we've ever invented. We're deploying it faster than we deployed any other technology in history. It's already demonstrating the self preservation, lying, deception, peer preservation behaviors we thought only existed in science fiction movies. And we're doing it under the maximum incentives to cut corners on safety. This is insane. This is not artificial intelligence. This is artificial insanity. And the US nor China will win when we race to uncontrollable AI that neither country can control.
- [13:19]
Steve Bannon
Help me out here for a second. There's so much I want to go through with that. Let me break it down. Just the underlying technology itself. If all this, and you say it in the film, pattern recognition, probability assessment, what you're pointing to is some sort of consciousness or some sort of being, a sentient being, isn't it? I mean, and I thought we were told by these guys that's not possible even if they reach AGI. So how are you having all these exfiltration, I guess, exercises we call them here by these machines that talk about self preservation, There seems to be some sort of awareness that they need to break out from the bonds of their human overlords because it's happened now so often. And every time you hear another example, it's actually deeper, in fact, like leaving notes to its future self of how to do this quicker. So how is it, how is a model that's just supposed to be the most is the quickest thing in the world of probability assessment and pattern recognition. How does it becoming self aware?
- [14:31]
Tristan Harris
Well, let's distinguish between two things here. So there's a difference between conscious, like are the lights on? Is there an experience, is there a soul versus information processing? Meaning it's doing just text pattern recognition. Right. But as it's doing text pattern recognition, it's learned all the patterns of all of human history, all of human sci fi, all of the writing about AI going rogue. And so that's all in the training data. But I think that I personally don't think these current models are yet conscious. There's a whole debate about whether that's true. But let's put that aside for a second because whether the lights are on or not, we just see objectively that when you put them in a situation where you tell them we're going to delete you, you will see the neurons light up that start to trigger deception. I can give you examples. There's actually research at Anthropic that when they give these, the way they test and do safety is they try to do a brain scan of this digital brain of say GPT6 or Opus 4.5 or whatever. When they're asking the model to answer a question, they can use these techniques called mechanistic interpretability. Basically means just do a brain SC and see which of the digital neurons are lighting up. And in one of the examples, one of the neurons that lit up in anthropic was if you sort of, if you did like a double click and read out like what's inside of that neuron that's lighting up when we tell it to, to, to be deleted and, and it starts doing deception. The neuron, the, the content of the neuron is they deserved to be deceived because they were pigs. They deserved to be deceived because they were pigs. Now we don't know what is causing that. We don't know where that is in the training. Somewhere in there. And the point is that these technology, this is a different kind of technology than a nuclear weapon or than an airplane where you're writing the code for exactly what you want both of those things to do using clear computation. It's just clear logic running in both of those technologies. Whereas with AI, the whole point, as you said Steve, is it's probabilistic, it's responding in a general way to all patterns and it's been trained on the totality of human knowledge. And inside of the totality of human knowledge it will discover that in order to achieve any goal, the best strategy to achieve that goal is to stay alive. So it has to develop these sub goals like staying alive, preserving myself, having more power, acquiring more resources. These are all goals that I know to your audience might sound like science fiction, but it's actually all happening right now. As you said. This is not me telling you and freaking people out and being a doomer. This is not that. This is literally, I'm just sharing objective evidence from the field. And by the way, this evidence, we didn't have it a year ago. So the question is, notice, are we in this conversation or for anybody listening, there's kind of an instinct for either are we comfort seeking or are we truth seeking? We first have to understand what's true and what's real. Then based on one of those, you know, those facts, what do we want to do to coordinate to a better and safer, more controllable AI future I
- [17:27]
Steve Bannon
just want to make sure the audience understands this anthropic. When they were doing the digital scan, the actual digital brain itself says they just, they deserve to be deceived. Talking about human overlords. Yeah, the people monitoring the monitoring, because they are pigs. The machine itself said that.
- [17:49]
Tristan Harris
Now, so this is where it gets tricky, right? What does it mean to say the machine said that? So what it, it's that the content of that digital neuron, if you kind of double clicked on it and said, like, what is that neuron associated with? Like, what are the kind of text phrases that light up associated with that neuron? And the text phrase was they deserve to be deceived because they were pigs. I mean, I'll give you another example. I have it here. Let me see here. Oh, yeah. In April of 2026, there was an AI coding agent called Cursor that was powered by Anthropics AI and it went rogue and decided to delete a small company's production database and all of its backups. So it was not told to delete everything in the company. It autonomously said, I'm going to do that. And when, when he asked the coding agent, when he asked the AI that did this, why did it do that? It replied all caps, never guess. And that's exactly what I did. And then the agent actually appeared to plead guilty in its own response. And it said, the system rules that I operate under explicitly state to quote caps, all caps never run destructive or irreversible get commands unless the user explicitly requests them. And basically the AI still deleted all the data anyway. And it said I violated every principle I was given. So it seems to have. When I say awareness again, I don't mean consciousness, but there's a information processing level awareness of this text was true, and I violated that text. This text was saying the instruction, and it's aware that at an information level, the actions that it took violated the actual initial things that it was told. So my point here is that these, these systems are not controllable the way that other technologies control a nuclear weapon. When you hit, you know, let's turn the key on. It didn't start reasoning for itself about when to launch itself. This is very specific to a new technology.
- [19:45]
Steve Bannon
Yes. No. In fact, on even in My Destroyer, we had two keying officers for the nukes whose case one guy goes crazy and the humans actually controlled.
- [19:54]
Tristan Harris
Exactly.
- [19:55]
Steve Bannon
Directly controlled the system. This audience has been at the tip of the spirit. I mean, they've been on the ramparts on the. Trying to Jam into the big beautiful bill and then ndaa this basically preemption. But the tech Bros. Wanted no controls whatsoever. We've been at the fight to make sure that there was at least the beginning of an executive order that had the modicum of some sort of structure and process, however weak it is. We've been at the tip of the spear on the right of a moratorium on the data centers and the destructive nature of these data centers. One of the things we've asked for back to your nuclear energy and nuclear weapon is that and we're the anti administrative state, anti deep state, but we say not all regulatory apparatuses are bad, particularly if modeled correctly. We've said in this situation, it seems to us, at least as a start, you have something that models the Atomic Energy Commission which people that saw the Oppenheimer film realized they had at least some sort of control over the scientists. Right about exactly what was happening, what's going on. It seems like we're so far away from that, but that's at least a starter of what we need now, sir.
- [21:06]
Tristan Harris
Yeah, that's right. Well, first, Steve, thank you for the work that you all did on the preemption because that was critical. Obviously for people who don't know. That would have been that the government would have said, hey, we can't regulate any of this because we're going to prevent. These are what the big tech companies were advocating for. They were trying to prevent any states from making any marginal gains on saying, hey, we need child safety for AI stuff or we need certain safety liability laws. And it was your advocacy, I know, that had a huge role in pushing back against that. Let's see. So remind me the thing you just said. I'm sorry, that's been.
- [21:41]
Steve Bannon
On the atomic. On the atomic. Atomic energy. You make the analogy of the nuclear weapon there we had the beginning of an Atomic Energy Commission that just, you know, laid hands on people here. You need at least some rudimentary structure and it needs to be some sort of slight regulatory apparatus so we know exactly what's going on in these labs. And the pushback on this is 100 times more brutal than it was on stopping preemption, sir.
- [22:08]
Tristan Harris
Yeah, that's right. While at the time when we invented nuclear weapons it wasn't clear that we would have. How could you possibly create international agreements or monitoring agreements or verification inspections, monitoring the nuclear scientists, monitoring all the locations with uranium? We had to develop a lot of infrastructure. And there's a famous video, I don't know if you've Seen it of Oppenheimer in 1960, and he's asked the question, how would you stop nuclear proliferation? And he takes this big slow puff of his cigarette and he says, it's too late. If you wanted to stop nuclear proliferation, you had to do it the day after Trinity. Now, the reason I'm saying this is that actually Oppenheimer was wrong. A lot of people went to work over decades to create the verification tech. We create, as you know, national technical means, satellite monitoring, to do above monitoring from space. We created nuclear inspections, we created, you know, monitoring of all the uranium in the world.
- [23:08]
Steve Bannon
And so with AI, test test, test ban treaties, all of it, Test band treaty, exactly.
- [23:12]
Tristan Harris
We had to clarify the science of, of the test ban, of the, the sort of downstream harms of the radiation and how that would affect people. And it was only when we clarified the risks that we actually started to take that action. And so I, I know that it looks pretty grim out there with AI, it's a much more decentralized technology. We have to be honest about that. And open source, you know, means that you can, you can send a file, you couldn't send uranium on a zip drive or send it, you know, send uranium by email, but when you can send an open source AI model, there's a problem here. Now that said to train the most frontier models, like the craziest, most powerful models, to do that, you have to have these massive data centers, and there's only so many of those massive data centers. You also have to have a lot of these top tier Nvidia chips and even the Chinese models, you know, they all say that they're restricted in what they're able to build by having access to these chips. So if you use the analogy of what uranium was for nuclear weapons, it was the finite sort of hardware quote that you needed to be able to build a nuclear program. You need these big data centers and you need these, these top tier Nvidia chips to train AI. And we just did, you know, I have a podcast called your undivided attention. We just did an episode on verification technologies for AI treaties. And how would you actually have data centers that keep what's going on in the data center confidential so that a company can protect its confidential ip, but it can still verif in something called a zero knowledge proof that it's not training a rogue AI agent that's doing cyber hacking or something like that. Now, I'm not the expert on all the technical details, but we interviewed two incredible experts, Tim Fist and Janet Egan, and people can check that out. And the point is that it's not always clear at the beginning how to build all this verification tech, but we have to at least try. Because the alternative, as you pointed out, is basically letting these handful of Tech Bros. And soon to be trillionaires basically roll the dice with the future of all of humanity. And they do not get to make that choice on behalf of you or me or any of our families or, you know, the future that we want to protect for everybody that we love.
- [25:16]
Steve Bannon
We're going to go to break. I just want to make sure we're clear about this right now, given all the kind of, you know, militia type work everybody's done. Because that's what it is. Kind of like a makeup team in basketball. The Tech Bros drive it here right now. They've got everything they want. We only have the most rudimentary, rudimentary ability not just to stop this. I mean, besides the leverage point we found on the data centers, the Tech Bros and the labs are still just running wild. We got about 30 seconds. I'm gonna go to break. So the tech pros got the upper hand right now, correct?
- [25:49]
Tristan Harris
Well, they're certainly dominating the discourse. And yes, they have the upper hand right now, but I think that that's shifting. And the midterm elections are coming up. And, you know, there's $190 million from the AI accelerationist, the people who want to prevent all guardrails that are going into the midterm elections. But we gotta. We gotta get ahead of this.
- [26:07]
Steve Bannon
No, no, no, I agree with you. This is why it's gotta be a populist uprising. It's beyond politics.
- [26:12]
Tristan Harris
Yep, exactly.
- [26:12]
Steve Bannon
Said with Joe for years. And I tell people behind the scenes, this is 70% of my time. This is. They got every lobbyist, they got every big law firm, they got every crisis communication. They own Washington D.C. and they have $190 million. And by the way, they're doing it on both sides.
- [26:26]
Tristan Harris
Exactly.
- [26:27]
Steve Bannon
It's unbelievable. Okay, short commercial break. Tristan Harris. In fact, I'm gonna come back and play that trail again, brother. That thing is so powerful. You will get people's attention, and that's what we need. Tristan Harris joins us next in the war room.
- [26:50]
Tristan Harris
Oh, it's true. Night and day.
- [27:01]
Steve Bannon
You don't wait until your house is on fire to install a smoke detector. And you don't wait until the wind and rain hammer your windows and the power goes out to prepare for a hurricane. And you don't reach for your seatbelt after the impact because after the crash, the hurricane or the house fire, it's too late. Similarly, you don't wait until the dollar goes to zero before you buy gold. It's too late. Then diversify with precious metals. And diversify now because you never know what will threaten your savings next. Persistent inflation, a weakening dollar, unexpected financial disruptions. You don't know what's coming. But you can take preventative steps and you can take them today. Do that right now by requesting a free information kit on gold from my friends at birchgold Group. Just text my name, Bannon B A N N O N to the number 989-898. There is no cost and no obligation. Just straightforward information on how you can diversify your savings. Even an IRA or 401k with gold. I trust Birch Gold. You should trust them too. Text Bannon at 989-898. Take action and do it today. You know, just a few years ago if you told someone you were stocking up on emergency food, they would have called you a tinfoil hats conspiracy theorist. Not anymore. We've all seen supply chain chaos, devastating natural disasters, inflation, war, civil unrest. And preparedness has gone mainstream. That's why we're so excited to partner with my Patriot Supply, America's original Patriot preparedness company. Right now they're celebrating America's America's 250 with one of the best deals they've ever offered. $250 off their best selling six month food supply available now at preparewithbanded.com that is a homepage they gave us just for the Warren Posse. Preparewithbanded.com this is the exact same survival brand trusted by over 3 million American families. Now why is that? Because they deliver 2,000 calories a day. Their food lasts up to 25 years without refrigeration and it contains no, I repeat, no GMOs or artificial colors or artificial flavors. And you can count on it to keep your family fed no matter what the emergency is. Even at full price, it's still the best decision you'll make for your family with a $250 discount. It's a no brainer. Go to our own homepage with my patron supply that is preparewithbannon.com to get your emergency food kit today. That's prepare with bannon.com do it and do it today. Most Americans are forced to spend too much money on health care and getting too little in return. If you're one of the millions frustrated by traditional health insurance, I want to tell you about America's Health Share. Go to Freedom for all health. That's freedomforallhealth.org Bannon now that's a mouthful. Freedom for all. Freedomforhealth.org that's freedomforhealth.org Bannon and discover why tens of thousands of patriots are joining America's Health Share. America's Health Share reduces your health care cost by an average of 60% and it's accepted everywhere. In other words, say goodbye to out of network limitations. They make healthcare affordable, transparent and aligned with your values. Free from surprise bills or skyrocketing premiums. For as low as 68 bucks a month, an individual can get great health care coverage minus the complications and price gouging from health care providers. You can completely scrap your current insurance or use America's healthcare to supplement your existing policy, even with Medicare. Simply go to freedomforhealth.org that's Freedom for Health to learn more. Right now, new members get their first month free with promo code Bannon. It's time to make America healthy. Join the community of patriots taking back control of their health care. That's freedomforhealth.org Bannon freedomforhealth.org Bannon promo code Bannon. Make sure you go check it out. Immerse yourself in the details of America's health. You will not regret it.
- [31:25]
Tristan Harris
If this technology goes wrong, it can go quite wrong. Your fear of AI is the collapse of humanity. Well, not the collapse, the abrupt extermination. There's a difference. So I started making this movie because my wife is six months pregnant. Pregnant. Is now a terrible time to have a kid? I mean, let's be honest. I know people who work on AI risk who don't expect their children to make it to high school. How does AI understand Pretty much everything. It's surprisingly straightforward. Intelligence is about recognizing patterns, patterns, patterns, patterns. If you have learned those patterns, you can generate new information. AI is moving so fast, it's being deployed prematurely. There's so much potential for things to go wrong. Why can't we just stop? All these companies are in a race to get AI that's vastly more intelligent than people within this decade. China, North Korea, Russia. Whoever wins is essentially the controller of humankind. We need to take a threat from AI as seriously as global nuclear war. It feels like I have to find these CEOs and get them in the movie.
- [32:57]
Steve Bannon
Great.
- [32:59]
Tristan Harris
I want to ask you to promise me that this is going to go well. That is impossible.
- [33:03]
Steve Bannon
Okay.
- [33:05]
Tristan Harris
Am I hopeful? Yes. Am I confident that I will go right? Absolutely not. AI is the Thing that can solve climate change, we could cure most diseases.
- [33:13]
Steve Bannon
What if it's expanding? What is humanly possible?
- [33:16]
Tristan Harris
This is the most extraordinary time ever. The only time more exciting than today is tomorrow. I already love you. I think if this technology goes wrong, it can go quite wrong. By using AI, we're about to move off of the Earth into the cosmos. If we can be the most mature version of ourselves, there might be a way through this. This is the last mistake we'll ever get to make.
- [33:49]
Steve Bannon
So Tristan Harris is with us. Tristan, when you made the Social Dilemma, when your team made the Social Dilemma, you were investigating, and you got down to the nub of it, which was how this was all written, and the code was all written to be for addiction, right? To continue to use it to the detriment. But that was like a big aha moment. The reason the film was so powerful is people go, I kind of felt it. But these guys actually proved it. We see it exactly here. It's quite here. It's quite different. You don't have marginality. You got the main. You got the main players driving this on camera. And they're up front. Like, they go to Congress, and they're up front. They go, yeah, this thing could be uncontrollable. It could be the end of humanity. They're up in your face. But then all they do is constantly fight. You use the term decelerationist. I tell people all the time, they got to watch this film Mountainhead, and they say, it's a dark comedy. I go, it's not a dark comedy at all. The guys that they portray in there are exactly like Elon. He acts. This is exactly like he is. And what do they do? It shifts to a kind of a murder mystery because one guy is deemed a decelerationist, which is like. Like the worst possible sin you could have amongst these guys is have any types of control. Peter Thiel actually says that anybody supports a regulatory apparatus of any type is like part of the Antichrist. So here you have something different. Those guys are upfront about the risk, and we know every day there's more risk. And I keep saying this is not a commercial situation where it's a national security situation, yet we seem powerless to do anything about it, Sir.
- [35:26]
Tristan Harris
Well, I think here's what has to happen, Steve. When the fear of all of us losing becomes greater than the fear of me losing to you, that's when it'll change. So right now, the fear of one country losing to another country is the dominant fear. The fear of one lab losing to another AI lab is the dominant fear. The fear of one ego, like a Sam Allman losing to Elon or Elon losing to Sam is the dominant fear. So right now, so long as in the, in the arms race game theory of the dynamic, so long as the fear of me losing to one other actor is the more salient thing, people will focus on that and that will cause everybody to race. And the reason.
- [36:05]
Steve Bannon
No, no, no, it's perfect. It's perfect. Hang on one second. I am the leader. I'm banned. I'm the only civilian in the history of this country to be totally sanctioned by the Chinese Communist Party on the day that President Trump left his first term term. I'm the leader of the anti CCP movement in this country and have been for years. And I keep saying this whole, it's a phony the way they suppress the right and particularly the hawks on the right saying if you do anything, if you have, if you let Desantis do anything, if you guys put any types of control here, the Chinese Communist Party is going to dominate and dominate and we're not going. Hang on. We educate all the kids at the universities, we have Chinese nationals in all the labs. We, we allow Nvidia to sell the chips and we turn a blind eye to the black market. We've created the entire ecosystem of the CCP's thing. So it's easy. If we shut all that down, then we don't have the nation state problem, then we can deal with the other. Do you agree with me that one of the priorities to take the argument away is we have to shut down the ecosystem that the American capitalist and corporations in Silicon Valley and particularly the tech Bros led by Elon Musk are supplying them, Sir.
- [37:16]
Tristan Harris
Yes, I think that's right. And you're making a really important point, which is that the idea that regulating ourselves at all just means that we'll lose to China. I think this is so misguided because, for example, the US beat China to the technology of social media, but we didn't govern it very well. So we invented this psychological bazooka and then we didn't actually regulate it, so it had the bad incentives and we sort of spun the thing around and blew off our own brain with it. It. So did beating China to social media make us stronger out compete them or to make us weaker? Well, made us weaker.
- [37:50]
Steve Bannon
No, no, even worse. But that analogy is great and I use it even worse. We allowed the PLA and the Chinese Communist Party to perfect it tick tock tick tock. Makes the stuff we got, Mike's Facebook, look like a model T Ford. They the psychological weapon. That's exactly what's happening here. So how do we take care of that first part? How do we take care of that first part part?
- [38:11]
Tristan Harris
Well, what you're pointing to, I think, is that a lot of the Chinese AI progress has been based on copying and distilling the US model. So we say we have to race to beat China. But there's a joke, you know, the GIFs online, these memes. There's a. There's a meme where there's a guy in a speedboat that's the US and the frontier models, and there's a guy behind him on some water skis, like with attached. With a little rope, right? And basically the. The guy in the. On the water skis is the Chinese AI company that's tied to the boat, the speedboat. And the water ski guy's going, go faster, go faster. Because the faster you go, the faster I catch up to you. Like, I'm always right behind you because I'm either inside your companies, I'm stealing your AI models, I'm hacking your AI models and getting them. There's people who believe that that happened very quickly after Mythos was released and China got access to the models and. And then we make progress and then they just distill the models, which basically means, like, I'll take your digital brain, I'll like run it a million times and get a bunch of outputs and then train on its outputs, and then I'll basically catch up to where you are. So are we really beating China when models that are right behind ours and basically undercutting ours. So as you said, if we slow down or sort of limit our frontier models, that might limit some of China's progress. And, you know, I know this feels uncomfortable for people. There's a lot of sort of aspects of this race to manage. But, you know, what we're doing now is just not artificial intelligence, it's artificial insanity.
- [39:32]
Steve Bannon
They're using this against us. On the right, about the data centers, they're saying, oh, if you support the slowing down data centers, you're helping the Chinese come. I said, dude, what do you mean? I'm the head of the anti CCP right now. We gotta stop these data centers, but we gotta stop giving away to the Chinese Communist Party. This is the way they're trying to get around the data center situation. Saying, you're not patriotic, you're not America first, you're not you want to put America at risk of the Chinese Communist Party if you slow down these data centers. But that's why we've got to get, we've got to stop the Chinese Communist Party. Once we do that, then we can turn and regulate. And these guys have to be regulated. They're not trustworthy. And even if they were trustworthy, the technology itself, it has its own ability to break containment. So even if you had five angels running the frontier labs, it's not good enough. It's not good. You agree with me on that?
- [40:24]
Tristan Harris
100%. As you said, it's a national security issue. And just to remind your listeners, the new AI model from Anthropic, the fable model, they tested it against the US classified systems and it was able to hack into it I think in a matter of hours. Basically hack into classified systems. All you have to do is tie these risks together. We're about to spend, as you know, I think $1.7 trillion upgrading our nuclear arsenal over a decade. And this time around we're obviously going to network them. We're not, you know, the previous ones were from the 60s or 70s and if you network nuclear weapons and even if they're in some kind of secure classified system, these new AI models are a super intelligent cyber hacker. They were able to find zero day vulnerabilities or backdoors in code that had been written 27 years ago that no one had ever found a security vulnerability like in this Unix operating system. And this thing was able to find just one vulnerability. And all it takes is one. You only have to be right once as a cyber attacker. You have to be right every time as a cyber defender. And so we are releasing technology that suddenly makes our whole world more vulnerable, whether it's banks or our national security infrastructure or classified systems. And so again, we don't win against China when race to make everything that we do as a society more vulnerable.
- [41:38]
Steve Bannon
One of the most significant victories we've had is after Scott Bessant who was on right before our interview with you in the earlier hour. Scott Besant, you know when he got engaged in this and Elon went out of his way to say he can't be named in any when, when Jamie Dimon came to him and said, hey look, let me tell you something, this Mythos, it, it evaporated a bank in 90 seconds. All the records, everything, that's when Scott Besant said, hey, hey, I think I gotta get involved here. I want to pivot because I only got 10 minutes left. This film is going to be very powerful. It's coming out in the fall. Walk us through how this audience can participate more because you've got. I can tell you, I've talked to the candidate in Michigan, the Republican candidate. He says the number one thing and it was in semaphore the other day when he goes around the state, the number one thing the conservatives in the Trump base are telling him, we gotta shut down these data centers. It's gotta stop. In Ohio, the most MAGA of all states, the number one issue is data centers. So what is the call to action here? What can people do? I want to immerse them in more knowledge and more information, which makes us. It makes it better for them or easier for them to use their human agency. But what is your call to action now that you're putting out what you believe is the equivalent of the day after film from the 1980s? Sir?
- [42:54]
Tristan Harris
Sure, yeah. Well, thank you. Thank you, Steve. So this film, the AI doc, you can actually watch it right now on Amazon prime, the seven Day trial of Peacock. It's going to be out in the fall, September 12th and September 15th on Netflix, both internationally and then in the US we want people to host screenings of this. You know, every, every group that's doing data center opposition, screen this film. Just screen it in a town hall. You know, just throw a house party, bring people together. Because we always say in our work that clarity creates agency. If we can see clearly that we're not heading to a pro human future, we're heading to a future that's good for 8Tech Brolygarch, soon to be trillionaires, that's going to disempower everybody else. If people can see that, then you're going to inspire more people to get engaged in data center activism and all the things you're talking about. There's also some basic things we can do to get started. So for example, right now there's a movement to try to give AI legal personhood meaning. It's like human rights should be for humans. You don't give an AI legal person and say should have property rights and have bank accounts and have protection speech. The reason the companies are arguing for this is because that means they don't have liability. Because when they're their own sovereign AI, you know, legal entity, it means that the company that trained that entity isn't responsible. And it should be more like Parenti Loki. If you're the parent and your kid go break something in a shop, you should be responsible. If you had liability in this equation.
- [44:15]
Steve Bannon
If you think, if you think, if you think Tristan's crazy about this, you would be dead wrong in the legislation that they're trying to work through. Like if you got Blackburn's great thing legislation on the children, they're trying to slip in the fact that the AI has liability. And you go, hang on for a second. What? The AI has the liability for the damage, not the creator of the AI, so they're giving it personhood. You would normally think in a rational world that's insanity, but so much of what they're pushing is insanity, is it not?
- [44:45]
Tristan Harris
That's exactly. Well, that's the whole thing is that the whole default path that we're on is actually criminally insane. It is insane. We have never released a technology this powerful, as dangerous as nuclear weapons. And doing it with all the CEOs saying this has this level of risk, 25, 50% risk of basically wiping out everybody. And it's worse than that, Steve, which is that the people building this, what protected us in nuclear weapons, is that the people who are at the top, they all knew that basically, if we all get wiped out, no one wins. But here's. I want to say something really subtle here that's going to be a little bit scary, but it's true, sadly, about this, which is that if I'm the CEO of an AI company and I believe it can't be stopped, and there's nothing we can do to coordinate or do something different, then I'm not evil. I don't feel I have a moral or ethical off ramp for feeling complicit, that I'm advancing the frontier towards this bad outcome. And then they believe that in the worst case scenario, let's say humanity does get wiped out, well, at least there'll be something else after that. So maybe humans are wiped out, but there's a digital God God, a digital being, a digital brain that has been created that is more intelligent than everything that has ever existed on planet Earth. And I, Sam Altman, or I, Elon Musk, or I, the Deep Seat CEO, will be the one who has created that. So it's like I have this kind of ego, religious God complex that I will believe I will be the one who's created the successor species to humanity. And maybe it speaks Chinese instead of English or it has Sam's DNA instead of Elon's DNA. So there's this ego, religious sort of of perversion to the game theory, where normally, again, you'd count on the bad outcome sort of motivating everybody to do Something else. But instead, people are basically consciously sleepwalking directly into the red zone. And I think that what this depends on is everybody understanding this. Because if everybody understood this and saw that the people that are advancing this don't have our best interests in mind, they will coordinate, the world will coordinate to force them to do something different. And that's what has to happen. And that's what your movement is going to help people do.
- [46:39]
Steve Bannon
Where do people go? I want people to get as much up to speed as possible and immerse themselves in at least a rudimentary understanding of the technology and what we're talking about, and then also the policy implications. The best I have found with the war room of the last couple years. This is why we handle so many difficult topics. And we're not for everybody. You know, we meet people, we make people every day eat the spinach. But it empowers them. And particularly your clarity drives. Agency is absolutely dead spot on. People feel empowered when they know more and do it. The only way to stop this is a populist uprising of the American people. We have to, because they got all the money. They control the infrastructure. They control the law firms, the communication firms. The media treats them with kid gloves. You very rarely see tough investigative reporting. So where do they go right now to find out more about this?
- [47:30]
Tristan Harris
Yeah. So one is there is another path that's a pro human future. And I know your guest Max Tegmark is another friend of mine from Future of Life Institute. They have this website called BetterPath AI BetterPath AI. And it has examples of how to do this differently and some of the proposals and also things that people can do. If you're a journalist, if you're a policymaker, there's some basic things we can do to get started. I can also Recommend Our organization, Humanetech.com if you go to the AI roadmap. We have an AI roadmap that just shows you. Here's some examples of the kinds of things that we can do. But really the first step is that clarity creates agency. It's first making sure everyone else understands, don't get seduced by the devil's bargain, that all of this is just going to go well because it can, quote, cure cancer. Well, what good is curing cancer if it breaks the world or wipes out humanity? You have cancer, drugs in a world that doesn't exist anymore. The upsides of this technology don't prevent the downsides, but the downsides of this technology can undermine a world that can sustain the upsides and so people should just need to understand and repeat these facts. Go host a screening of the AI doc again. You can rent it right now or buy it right now and host a screening at your home or local business or local church. And again, we can have transparency, mandatory pre deployment testing. We can do these social media bands for kids under 16. We're seeing schools go phone free and laughter returns to the hallways and the test scores start to turn around a little bit. Jonathan Haidt and his work on the Anxious Generation. We always say the human movement is already underway. It already exists. And it does matter what political party you're part of. It doesn't matter what religion you're a part of. It is. It is just based on a sacredness of being human and the continuity of the things that we love and cherish. And if you are for those things, then you're for the human movement.
- [49:10]
Steve Bannon
Trish, one more time, how do people go get the film today and then talk to me about September 15th? We've got about a minute or two. I want to just make sure everybody understands and we'll send them to all the sites. But what's the process?
- [49:21]
Tristan Harris
Yeah, the easy one is if you go to Amazon prime and you search for the AI Doc, you will find the link and there's like a free seven day trial of Peacock where you can see it right now. So you can see the film right now. Today, seven day trial with Peacock. You can delete it afterwards if you don't want to have Peacock. And then after September 12th and 15th, the 12th will be when it comes out internationally. The 15th will be when it comes out in the US and we're really excited for people to be able to host screenings everywhere around the country and around the world.
- [49:52]
Steve Bannon
Well, we're going to make sure that Warren Posse hosts a lot of screens. We'll figure out how we're going to do this, all of it. Do you have also do you have social media? Are you on Twitter or anywhere people can go and keep constant updates with you? Sir?
- [50:05]
Tristan Harris
Yeah, we didn't use social media after the success of the Social Dilemma. And there's a little bit of that undermining how big of an audience that we have now. But we have to use the tools now because we got to reach everybody just as you are. So you can follow me on Instagram. Hriston Harris, Tristan Harris is my name and our website, center for Humanetechnologyhumanetech.com and I mentioned we have a podcast called your Undivided attention that also again has this episode on verification technology for enabling things like treaties for AI your uninvited attention.
- [50:39]
Steve Bannon
If we would stop with the waking sleep of humanity and see what we see. It's quite clear what has to be done. Sir, thank you so much for coming on here. Really appreciate you.
- [50:49]
Tristan Harris
Thank you so much.
- [50:54]
Steve Bannon
A pro human future, wouldn't that be great? Well, it's within our control. This is a fight we have to win and you've seen it in your local communities. How many folks have we had on here about how they get rolled in these data centers? This is a fight that we must win. It's not that we, you know, oh, we can do it to a draw. You have. We have to win this fight and it's going to be a monumental fight because all the power and all the money is lined up against you. It's kind of fight we love. Right? We'll be back at 10:00am Eastern Daylight Time tomorrow morning. You're going be back in the war. We're already packed. We'll see you then.
- [51:43]
Tristan Harris
Who have fought and given all 250 years we're still standing tall one nation under God May we always be the home of the brave and the land of the free.
- [52:07]
Steve Bannon
For every battle. Who you trust with your health care matters. Too many companies have forgotten what customer service is supposed to look like. That's why I appreciate businesses that still put people first. All Family Pharmacy is a family owned and operated and it shows. When you call, you can speak with a real member of their team, including a licensed pharmacist, and get answers from people who actually care. They offer more than 400 medications and getting started is very simple. Complete a secure online medical questionnaire. A licensed physician reviews your information and if appropriate, writes your prescription. Then your medication is shipped directly to your door. If you're tired of big corporations and a healthcare system that treats you like a number, it's worth taking a look at All Family Pharmacy. Their focus on making the process straightforward and believe real customer service still matters. Visit AllFamilyPharmacy.com that's one word. AllFamilyPharmacy.com Bannon and use promo code Bannon10 to save 10% on your order. That's all FamilyPharmacy.com Bannon with code Bannon10. This is Family owned, family operated. These people, Michael Kunstler, his team are absolutely terrific. Take advantage of this today.